A variable latency elastic buffer comprises a plurality of memory locations in which to hold data. A write and read pointer may point to respective write and read addresses of the plurality of locations in which to write and read data. A controller may hold or increment the address of the read pointer upon determining that the amount of data within the buffer differs from a nominal fill level. In a particular embodiment, initialization circuitry may be operable to initialize the read and write addresses of the respective pointers responsive to an initialization request. The read and write addresses may differ from one another by an offset value equal to a value programmed for the nominal value.
|
1. A buffer circuit comprising:
a plurality of memory locations to hold data;
a read pointer to point a read address of the plurality of memory locations from which to read output data;
a write pointer to point a write address of the plurality of memory locations in which to write input data;
the read pointer and the write pointer operable responsive to a read clock signal and a write clock signal, respectively, to sequence the read address and the write address, respectively, across an address space of the plurality of memory locations;
a control register to store a nominal level for the buffer circuit;
the control register being programmable for receiving the nominal level for setting latency of the read pointer;
the latency of the read pointer being variable responsive to a fill level of the plurality of memory locations relative to the nominal level; and
a controller to affect operation of the read pointer dependent on the write address of the write pointer, the read address of the read pointer, and the nominal level,
wherein the controller is operable to:
determine an amount of data in the buffer circuit based on a difference between the write address and the read address,
cause the read pointer to increase the read address in advance of the read clock signal responsive to the amount of data being greater than the nominal level, and
wherein the controller is configured to adjust the read pointer independently of the read clock signal.
2. The buffer circuit of
3. The buffer circuit of
4. The buffer circuit of
configure the write pointer with a predetermined start address for the write address; and
configure the read pointer with a beginning value based on the predetermined start address for the write address and the nominal level.
5. The buffer circuit of
|
This application is related to co-pending U.S. patent application Ser. Nos. 10/082,490, 10/090,250, and 10/234,978, each commonly owned by the assignee of the present application and incorporated herein by reference in its entirety.
The present disclosure relates to data communication and more particularly to buffer circuits and their methods of operation.
As communication technologies evolve, data transfer speeds increase along with the amounts of data. These parameters continually push the demands of the data processing devices to provide for circuits that may handle more data with faster throughput.
At the same time, the semiconductor industry has been developing integrated circuits of greater density and smaller size. With these increased levels of integration and functionality within the smaller area devices, a greater premium exists on the availability of signal interfacing.
To assist these I/O needs, transceiver-multiplexing circuits may be used to replace parallel data lines with higher speed serial interfacing. Using a transceiver-multiplexing circuit, a receiver may receive a high-speed serial data sequence and then convert it into parallel data of a slower clock rate. Likewise, data that is to be sent out may be converted from a parallel format to a serial data stream.
To further enhance the data handling capability, the serial-to-parallel and parallel-to-serial transceivers may be disposed in parallel relationships on a given data-handling device. This structure may enhance the efficiency of the more highly integrated data processing applications of, e.g., encoded data processors or programmable circuits of a field programmable gate array (FPGA).
When receiving data of an external serial data stream, the transceivers may accommodate input data transfer speeds as high as 3-10 gigabits per second, or even higher. This speed of operation, in some instances, may be much greater than the internal operating speed of an embedded data processor/system. But use of the serial-to-parallel transceivers and conversion of the serial data into parallel data may effectively reduce the high-speed data rate of the serial stream to a slower rate.
For example, a serial data stream may be encoded with, e.g., “non-return to zero” encoding of a known protocol to assist clock recovery processes. It may arrive with a 3 gigabit per second (Gbs) data rate. Assuming an 8 bit/10 bit (non-return to zero) structure, a decode and clock recovery process may replace every 10 bits of the input stream with 8 bit recovered data and may establish a corresponding clock rate of about 2.4 Gbs. Subsequent serial-to-parallel conversion to 16 bits of parallel data may further reduce the effective data/clock rate from 2.4 Gbs to about 150 megabits per second. This lower data rate may be more easily accommodated by most off-the-shelf and/or “library” data processing circuits, systems and embedded components.
From the above example, the effective I/O pad count for an integrated circuit may be reduced by a factor of 16-to-1, or more commonly 8-to-1 when the serial data is a differential signal. That is, the pin count for the integrated circuit may be reduced proportionately in accordance with the conversion factor of the transceiver.
When working with fast input data streams, some communication systems may recover an input clock from transitions of the received input data (rather than accompany the data signal with an associated clock signal). Using known data encoding protocols, the data to be delivered may be encoded to assure a sufficient number of signal transitions that may support clock recovery. For example, eight bits of data may be encoded into a 10 bit sequence. The encoding may add additional transitions to facilitate AC-coupling and clock recovery. Thus, for example, a hexadecimal data “FF” (1111 1111), which might otherwise present eight consecutive high level conditions for serial propagation, may be alternatively encoded to a 10-bit sequence with additional zero states (101011 0001). The encoded 10-bit serial data stream encoded by such known encoding protocols may also help guarantee a DC-balance, which may be helpful when using AC coupling.
It may be further noted that the different data channels may present data to a receiver with different timing relationships. In other words, the data of one channel may be offset or “skewed” relative to the data of the other channels. Additionally, the frequency from which the data originated may not be coherently related to the internal operating frequency of the receiver. In other words, the data may originate from a transmitter having a clock frequency that may be different from that of the receiver and that may also drift over time.
Because the data of the separate parallel communication channels may not be aligned relative to each other at the receiver, and because the frequency associated with the received data may not be coherent to the internal operating frequency of the receiver; buffers may be required to buffer and align the data of the different channels. Additionally, the buffers may correlate the data that is recovered to an internal operating frequency of the receiver.
In accordance with an embodiment of the present invention, a variable latency elastic buffer may comprise a plurality of memory locations to hold data. A write pointer is operable to point a write address of the plurality of memory locations in which to write input data. A read pointer may point a read address of the plurality of memory locations from which data may be read. The read and the write pointers respond to respective read and write clocks for incrementing their addresses across the address space of the plurality of memory locations. A control register may receive a nominal fill value to define the nominal amount of data to be held in the buffer. A controller may adjust the address of the read pointer responsive to a determination that the amount of data in the buffer differs from the nominal fill value presented by the control register.
In accordance with a further embodiment, the controller may temporarily hold or decrement the read address of the read pointer when determining an amount of data in the buffer less than the nominal value. Alternatively, the controller may advance the read address of the read pointer upon determining the amount of data to be greater than the nominal value.
In accordance with a further embodiment, the controller may first determine if the amount of data differs from the nominal amount by at least a predetermined threshold before enabling the hold or advancement of the read address.
In accordance with another embodiment of the present invention, a channel alignment circuit may comprise a plurality of buffers to buffer and synchronize data of a plurality of data channels. Correction logic may pad or omit data from at least one of the buffers upon determining the amount of data therein to differ from a threshold value of a control register. In a further embodiment, a variable latency register may receive and present a nominal fill value for use as the control value of the control register. A comparator may then determine whether the magnitude of the difference between the amount of data in the buffer and the nominal fill value is sufficient to enable padding or omission of data for clock correction.
In a further embodiment, a generator may generate a threshold value based on the nominal fill value. Clock correction may then be triggered when the amount of data in the buffer differs from the nominal fill level by an amount greater than the threshold.
In accordance with a further embodiment, the variable latency register may be defined at least in part by one of the group consisting of a read-write memory, a content addressable memory, a look up table, a programmable fabric and an increment decrement register.
In accordance with a further embodiment, initialization circuitry may initialize the beginning read and write addresses of the respective pointers. The pointers may be initialized with an offset therebetween that is based upon the nominal fill level.
In yet another embodiment, a parameter for the packet length of data to be processed by the buffer may be determined and used to define the value for the nominal fill. In another embodiment, the value for the nominal fill may be defined dependent upon at least one of an identified protocol, a maximum skew between data blocks of two separate channels, and a difference between read and write frequencies.
In the following description, numerous specific details may be set forth to provide an understanding of exemplary embodiments of the present invention. It will be understood, however, that alternative embodiments may comprise sub-combinations of the disclosed examples.
Additionally, readily established circuits and procedures of the exemplary embodiments may be disclosed in simplified form (e.g., simplified block diagrams and/or simplified description) to avoid obscuring an understanding of the embodiments with excess detail. Likewise, to aid a clear and precise disclosure, description of known processes—e.g., triggering, clocking, state-machine, programming procedures—may similarly be simplified where persons of ordinary skill in this art can readily understand their structure and operations by way of the drawings and disclosure.
Referencing
For example, further referencing
A user may configure a plurality of circuits in programmable fabric 106. Some of these circuits may also communicate with transceivers 102-104. For example, further referencing
Referencing
Some of the configuration memory cells may be used to configure programmable fabric 106, while other configuration memory cells may be used to configure the transceivers (e.g., 102-104). For example, configurations memory cells in blocks 216 and 218 may be used to configure two separate transceivers, such as transceivers 103 and 104 in
Some memory cells may define logic block functions and a variety of interconnections to implement desired logic designs in programmable fabric. Other memory cells, in exemplary embodiments of the present invention, may be used for purposes of programming performance of the transceiver. Some of the memory cells associated with the transceiver may further be used to modify routing of data in the transceiver, thereby including or excluding a sub-component of the transceiver to work with the data.
In a particular embodiment, a portion of the transceiver may be configured via data of the configuration memory cells and another portion may be configured via control values established by logic in the programmable fabric. In an alternative embodiment, the transceiver may be controlled entirely by the configuration memory cells. Likewise, in further embodiments, it may be controlled entirely via the configured logic.
Referencing
A plurality of clock signals (shown collectively along signal line 140) may support clocking of data at transmit buffer 162 and receive buffer 172. Transceiver 130 may further comprise cyclic redundancy code (CRC) circuitry 186 of known provisions for operating a CRC status signal 142. Additionally, synchronization status circuitry 188 may be operable to provide synchronization signal 144 dependent on whether or not synchronization has been established or lost.
In one embodiment, the width of the data paths 136 and 138 may be independently configurable and selected to be 1, 2, or 4 bytes. In other embodiments, the data paths 136 and 138 may have other widths.
Addressing the transmitter side 150 of transceiver 130, data of data path 136 may be selectably (or optionally) processed by a CRC generator 152. This CRC generator may compute and insert known CRC, such as a 32-bit CRC, into data packets for transmission. Because different protocols may handle data in different ways, CRC generator 152 may recognize data packet boundaries by which to perform CRC computation for appropriate sets of data.
The resultant CRC data may then be delivered to encoder 154. In one embodiment, encoder 154 may comprise a known 8B/10B encoder. For example, it may use the 8B/10B code having 256 data characters and 12 control characters that is used in the Gigabit Ethernet, XAUI, Fibre Channel, and InfiniBand protocols. The encoder may accept 8 bits of data along with a K-character signal for a total of 9 bits per character applied. If the K-character signal is “High”, the data will be encoded into one of the 12 possible K-characters available in the 8B/10B code. If the K-character input is “Low”, the 8 bits will be encoded as standard data. If the K-character input is “High”, and a user applies other than one of the pre-assigned combinations, an error signal can be generated.
The encoded data may then be delivered to a transmit FIFO buffer 156. This buffer may interface encoder 154 to serializer 158. The encoder 154 may be controlled by a clock of the internal system on line 140. The serializer may be controlled by a reference clock generated by a transmit clock generator 160. These two clock signals may be frequency locked. Although frequency locked, the two clock signals may have different relative phase relationships. Accordingly, transmit FIFO buffer 156 may be operable to absorb phase differences between the two frequency-locked clock signals.
The data of FIFO buffer 156 may be delivered to serializer 158, which may multiplex the parallel digital data of FIFO 156 into a serial bit stream for transmission over a serial link. The serial bit stream may be output by transmit buffer 162, which may drive the serial bit stream onto a pair of differential lines of output port 132.
Turning now to the receiver side 170, transceiver 130 may further comprise a receiver clock generator 176 that may generate a reference clock signal to operate certain parts of the receiver. Receive buffer 172 may accept and buffer the serial data received from differential lines of input port 134. The buffered data signal may then be fed to a clock-data recovery block 173, which may recover a clock signal from transitions of the serial input signal. The clock recovered may comprise a frequency and phase based on that of the incoming serial data. This recovered clock may then be used to time the reception of the data within the deserializer 174, decoder 178 and elastic buffer 182.
Deserializer 174 may convert data of a serial bit format into parallel digital data. In a further embodiment, deserializer 174 may also perform comma detection. In some decoding algorithms, (such as the 8B/10B encoding), a “comma” may be used as a distinguishable pattern by which to assure determination of the byte boundaries within the serial data stream. For example, two comma patterns on known 8B/10B decoding may comprise a comma “plus” and a comma “minus”. Detection of a comma may then define the byte alignments within the received serial bit stream. In one embodiment, programmable cells may establish control signals to control whether the comma detection circuit is to realign the byte boundaries on comma plus, comma minus, both, or neither.
In exemplary operation, decoder 178 may decode digital data that has been previously encoded. When enabled, it may decode and, in accordance with its selected configuration, raise a synchronous “comma” flag (as a status bit to be attached to each received byte at the transceiver's programmable fabric interface) on comma plus only, comma minus only, both, or neither. In further embodiments, it may set this flag for valid commas only.
Upon leaving decoder 178, the decoded data may be sent to an elastic buffer 182. Elastic buffer may perform channel bonding and clock correction as driven by clock correction controller 184. Elastic buffer 182 may be configurable for various configuration options such as:
Configuration options might also include choice of channel bonding modes, the selection of a selectable number of channel bonding sequences, establishing a selectable length for matching a selectable byte value (8-bit or 10 bit), and similar provisions for clock correction sequences.
After re-synchronization by the elastic buffer, the data may then be delivered to an internal system, such as the programmable fabric, across data path 138. In one embodiment, the data width of data path 138 may be configurably selected to be 1, 2, or 4 bytes.
In further embodiments of the transceiver, elastic buffer 182 may be optionally connected to a known CRC verification block 186. This block may, for example, verify that the commonly used 32-bit cyclic redundant code (CRC) appears at the end of the received data packets. A signal may then be delivered to the internal system on line 142 to indicate the CRC verification.
Transceiver 130 may further comprise a loss of synchronization detector 188. It may interpret outputs of comma detector 174, decoder 178, and elastic buffer 182 to determine whether the incoming bytes of the data stream are in synchronization. A signal may then be delivered to the programmable fabric on line 144 to report the synchronization status.
Referring back to
Referencing
The array of memory locations may be driven and operable as a FIFO (First In First Out) buffer. In this mode, the last addressable memory location (e.g. A63) may be treated as addressable in contiguous relationship with the buffer's first memory location (A00). When receiving data, input data received, e.g., on input line 422 may be written into a location of the array as specified by the write address of write pointer 440. The write operation may be in synchronous relationship with respect to a transition of a recovered write clock, e.g. of clock generator 430. When the recovered write-clock signal on clock line 432 transitions, a data block may be written into a memory location as specified by the write pointer 440. Write pointer 440 may likewise increment its address. Responsive to the next transition of the write clock, the next consecutive data block from input line 422 may be written into the next contiguous memory location of the buffer. Upon writing data into the last addressable location, the next data block may be written into the first addressable memory location.
When sending data out from FIFO buffer 420, a transition of a read clock signal (on line 452 of generator 450) may trigger the output of data from a location of the buffer as addressed by read pointer 460. The transitions of read clock 450 may further increment the read address of the read pointer to the next consecutive address of the buffer. Again, upon reaching the last address, the read pointer may wrap back to the first address of the buffer.
Typically, the write address of the write pointer may lead the read address of the read pointer by an offset equal to, e.g., half the size of the buffer. For example, if the buffer comprises 64 addressable memory locations, a write address of 52 would establish a read address of 20 (=52−64/2).
In operation, the elastic buffer may absorb mismatches between the write and read clocks 430 and 450 respectively. In other words, data may be clocked into the FIFO buffer 420 by a write clock 430 of a frequency different from that of the read clock 450 used to output the data. To accommodate the difference between the clocks, elastic buffer 400 may adjust the data output in order to avoid buffer under/overflows. For example, if read clock 450 is faster than the recovered write clock 430, then FIFO buffer 420 may be emptying. To avoid emptying of the FIFO buffer, data of a particular block (e.g., a single byte or a sequence of two or four bytes) may be output multiple times. For instance, the read pointer may be driven so as to hold or decrement its value for at least one clocking transition of the read clock. As a result, the output data may be padded with redundant data and the buffer may restore or replenish its fill level.
Conversely, if the read clock 450 is slower than the recovered write clock 430, then the FIFO buffer may be seen to fill. To avoid overflow, the address of the read pointer may be advanced, which in turn may omit data of the buffer for output. By performing such padding/omission of data, the buffer may be described as “elastic” in keeping a fill level of the memory at a given level with “clock corrections” so as to avoid emptying or overflow.
As used herein, the padding/omission of data may be referenced more generally as “clock correction” or “rate matching”. By such clock correction, a data block of an original input stream may be omitted, or a data block added. In this embodiment, the data byte may comprise a single byte or a sequence of bytes (e.g., of two or four bytes). Additionally, the data adjustment may be aligned with a region of the data stream where the loss or addition of data might be of nominal affect. For example, the correction may be timed to coincide with an identified idle or gap sequence between data packets.
Further referencing
In a particular embodiment, the control circuitry may be viewed as disabling the clock signal from read clock 450 as presented to read pointer 460 so as to hold the read address. Accordingly, the output signal may receive redundant data that may correspond to a previous block or sequence output.
In a further embodiment, known protocol or sequence detection circuits (not shown) may detect a given sequence or gap between data packets. It may then generate a correction enable signal to drive enable line 472 of the read pointer control circuitry 470A. Activation of the enable line 472 may be operable to enable adjustment of the address of the read pointer coincident with a given sequence of the received data.
In the exemplary embodiment schematically illustrated in
In yet another embodiment, a flip-flop of the read pointer may have a clock enable input that may be driven dependent on the comparison. In a further embodiment, the logic circuitry of the read pointer may decrement the read address by a magnitude of one or by another decrement amount.
Referencing
Adder 716 may receive the selected value from multiplexer output 711 and may add this to the existing value 718 of the read pointer. Adder 716 may then present its output 720 to the read pointer as the next read pointer address—RD PNTR(N+1). The new address may be captured by the read pointer 460′ responsive to the next clock event of read clock 450.
In an alternative embodiment, referencing
In yet a further embodiment, this functionality of the read pointer control circuitry might be implemented in known fashion by a state-machine and/or a microprocessor programmed for performing these functionally illustrated operations.
Turning back with reference to
In further embodiments, logic of the read pointer may be configured so as to enable an increment of two or four or other amounts when enabled and upon receiving a clocking event of read clock 450.
In a particular embodiment, referencing
Again, similarly as described above with reference to
It may be further understood, that the two different address “hold/decrement” and “increment” control circuits of, e.g.,
In accordance with such embodiment, further referencing
In accordance with another embodiment of the present invention, elastic buffer 400 of
Referencing
Further referencing
For example, nominal fill control register 474 may be initially programmed with a value of ten. During initialization, the write pointer may receive a write address of 30. Subtracting ten from the write address, the initialization module 485 may then program read pointer 460 with a read address of 20. With this configuration, the FIFO buffer may operate with a latency of approximately 10 cycles, which may be proportionate to the amount of time in which data may remain in the FIFO buffer after being written therein.
A larger magnitude for the nominal fill of buffer 420 may allow elastic buffer 400 to accommodate larger differences between frequencies of the read and write clocks while still being able to maintain appropriate clock corrections. Similarly, the larger magnitude may allow it to accommodate greater skew between the data of separate data channels where multiple elastic buffers may be configured to operate in parallel. For applications of large frequency differences or large skew between parallel channels, a high latency and buffer size may be required. But the high latency may be excessively large for other types of data handling applications where faster throughput may be desired.
In accordance with an embodiment of the present invention, fill control register 474 may be programmable and may receive a user defined level so that the latency of the elastic buffer may be set in accordance with demands previously determined. Accordingly, a user may set a value for the control level for establishing a nominal fill level and elasticity of the elastic buffer. The value, in turn, could be based upon a pre-determined data protocol, expected packet lengths, anticipated clock differences, data skew, etc.
In an exemplary method of operation 500, referencing
In operation, further referencing
Then, in accordance with one embodiment, the amount of data may be determined (step 530) by subtracting the address of the read pointer from the address of the write pointer. This amount determined may then be compared (step 540) to a previously defined control level or levels. If the amount is determined to be greater (branch 542) than the control level (e.g., a maximum fill value), then control may transition to a data omission clock correction (step 550B). Should the comparison (step 540) determine the amount to be less (branch 544) than the control level (e.g., a minimum fill value), then control may transition to clock correction (step 550A) for padding data. After executing clock correction (by either padding or omitting data, as in steps 550A and 550B, respectively), the buffer returns to normal operation (step 548), and normal operation continues until the next opportunity to execute clock corrections.
Continuing with further reference to
In accordance with a further embodiment of the present invention, referencing
In accordance with one embodiment of the present invention, the threshold may be obtained from a threshold control register. Additionally, and in accordance with a further embodiment, during initialization, an initialization module may program the threshold control register with a value based upon the nominal fill level. For example, the threshold may be assigned a value equal to ten percent of the nominal fill. Accordingly, if the nominal fill level should be programmed with a value of ten, then the threshold control register would be configured with a value of one.
In accordance with a further embodiment of the present invention, the upper significant bits of the nominal fill control register may be used to establish the threshold level at the threshold input 473. For example, a nominal fill level of eight may establish a threshold level of four (by stripping the least significant bit). Or alternatively, a nominal fill level of 16 may establish a threshold level of four (should the lower two significant bits of the nominal level be stripped for establishing the threshold).
In yet a further embodiment of the present invention, the threshold value may be selected as one of a plurality of pre-determined levels dependent upon the magnitude of the nominal fill. Accordingly, the magnitude of the nominal fill level may be used to drive a multiplexer to select one of the plurality of different threshold levels. Finally, in further embodiments, the threshold level may be obtained directly from a pre-programmed read only memory (PROM) 202, such as that of
In yet a further embodiment of the present invention, an initialization routine, e.g., of a processor, may determine a nominal fill level for the nominal control register 474 by one of a variety of performance parameters of the elastic buffer. In one embodiment, an initialization algorithm may determine (step 505 of
Further, an anticipated skew may also impact the amount of buffering required within the variable latency elastic buffer. This performance parameter might also be incorporated by the initialization module to establish the nominal fill level.
It will be understood that any one of the above performance parameters alone, or in combination, may be used as the basis by which to calculate the nominal fill level during an initialization.
Returning with reference to
Referencing
At a receiver, received data sequences of buffers 612b, 613b, 614b, 615b of the four separate channels may not be aligned relative to the sequential arrangements of each other. For example, the receiver sampling clock 620 of the buffers may capture data of the second buffer 613b that is not aligned relative to the others. This misalignment may result from a variation of transmission delays, especially if the data streams may be routed through repeaters. In this particular example, further referencing
Through proper realignment, or channel bonding, the read address of the various buffer channels 612c, 613c, 614c, 615c may be properly interrelated by their read pointers and positioned to the desired buffer locations so as to capture the correct data bytes with the read clock.
In a particular embodiment, a special byte sequence may be positioned into the data stream of the various buffers. For example, “P” bytes may be designated as a special character for purposes of re-alignment. Accordingly, each receiving transceiver may recognize the “P character” and remember its location within the buffer. One of the transceivers may be designated as a master for instructing all of the other slave transceivers to re-align their channel for bonding to the designated bonding character “P” (or to some location relative to the channel bonding character). After this re-alignment operation, the words may be properly aligned: for example, RRRR, SSSS, TTTT, etc.
In accordance with a further embodiment of the present invention, with reference to
Again, in a particular embodiment, a comparator may determine when the amount of data in the buffer differs from the nominal fill level. Further, the value for the nominal fill may be based on a latency value presented by latency register 1095. Correction logic may be enabled to pad data for output responsive to determining an amount of data in a buffer to be less than the nominal fill value; or it may be enabled to omit data for output responsive to determining that the amount of the date in the buffer is greater than the nominal fill value.
In the above description, correction processes and embodiments have been disclosed that may provide adjustments to a read pointer. It will be understood that alternative embodiments may adjust the write pointer. For example, a query circuit may determine the presence of a gap between packets, a coding region, etc. A control circuit for the write pointer may then halt advancement of, or decrement, the write pointer, based upon the signal from the query circuit and determinations of the fill level of the buffer. For this alternative embodiment, when the level of the buffer exceeds the nominal fill, or other threshold, the write buffer control circuit may wait for enablement from the query circuit and then act to re-write previously written location(s).
While certain exemplary features of the embodiments of the invention have been illustrated and described herein, many modifications, substitutions, changes and equivalents may now occur to those skilled in the art. It is, therefore, to be understood that the appended claims are intended to cover all such embodiments and changes as fall within the spirit of the invention.
Cory, Warren E., Kryzak, Joseph Neil
Patent | Priority | Assignee | Title |
10216219, | Dec 07 2009 | Altera Corporation | Multi-protocol configurable transceiver including configurable deskew in an integrated circuit |
10289786, | Jun 27 2017 | XILINX, Inc. | Circuit design transformation for automatic latency reduction |
10484165, | Dec 19 2017 | STMicroelectronics International N.V.; STMICROELECTRONICS INTERNATIONAL N V | Latency buffer circuit with adaptable time shift |
11302370, | Apr 13 2020 | Fujitsu Limited | Semiconductor apparatus and synchronization method |
11361795, | Feb 28 2018 | SONY EUROPE LIMITED; Sony Corporation | Data buffer and data buffer control |
7787498, | Jan 26 2007 | Futurewei Technologies, Inc. | Closed-loop clock synchronization |
7809027, | Sep 25 2006 | Futurewei Technologies, Inc. | Network clock synchronization floating window and window delineation |
7813271, | Sep 25 2006 | Futurewei Technologies, Inc. | Aggregated link traffic protection |
7844762, | Feb 24 2006 | Lattice Semiconductor Corporation | Parallel interface bus to communicate video data encoded for serial data links |
7925814, | May 09 2007 | CHAOLOGIX, INC | Dynamically configurable high speed interconnect using a nonlinear element |
7961751, | Sep 25 2006 | Futurewei Technologies, Inc. | Multiplexed data stream timeslot map |
7986700, | Sep 25 2006 | Futurewei Technologies, Inc. | Multiplexed data stream circuit architecture |
8144160, | Feb 16 2007 | ASPEED TECHNOLOGY INC | Methods and apparatus for non-intrusive capturing of frame buffer memory information for remote display |
8289962, | Sep 25 2006 | Futurewei Technologies, Inc. | Multi-component compatible data architecture |
8295310, | Sep 25 2006 | Futurewei Technologies, Inc. | Inter-packet gap network clock synchronization |
8340101, | Sep 25 2006 | Futurewei Technologies, Inc. | Multiplexed data stream payload format |
8375115, | Feb 16 2007 | ASPEED TECHNOLOGY INC | Methods, apparatus, and systems for integrated management, graphics and I/O control of server systems |
8401010, | Sep 25 2006 | Futurewei Technologies, Inc. | Multi-component compatible data architecture |
8494009, | Sep 25 2006 | Futurewei Technologies, Inc. | Network clock synchronization timestamp |
8532094, | Sep 25 2006 | FUTUREWEI TECHNOLOGIES, INC | Multi-network compatible data architecture |
8588209, | Sep 25 2006 | Futurewei Technologies, Inc. | Multi-network compatible data architecture |
8605757, | Jan 26 2007 | FUTUREWEI TECHNOLOGIES, INC | Closed-loop clock synchronization |
8660152, | Sep 25 2006 | Futurewei Technologies, Inc. | Multi-frame network clock synchronization |
8837492, | Sep 25 2006 | FUTUREWEI TECHNOLOGIES, INC | Multiplexed data stream circuit architecture |
8838856, | Feb 16 2007 | ASPEED TECHNOLOGY INC | Virtual universal asynchronous receiver transmitter for server systems |
8976796, | Sep 25 2006 | Futurewei Technologies, Inc. | Bandwidth reuse in multiplexed data stream |
8982912, | Sep 25 2006 | FUTUREWEI TECHNOLOGIES, INC | Inter-packet gap network clock synchronization |
9019996, | Sep 25 2006 | FUTUREWEI TECHNOLOGIES, INC | Network clock synchronization floating window and window delineation |
9024878, | Feb 16 2007 | ASPEED TECHNOLOGY INC | Hardware cursor snooping |
9106439, | Sep 25 2006 | Futurewei Technologies, Inc.; FUTUREWEI TECHNOLOGIES, INC | System for TDM data transport over Ethernet interfaces |
9106504, | Oct 16 2013 | Altera Corporation | Methods and apparatus for aligning signals in transceiver circuitry |
9443618, | Dec 27 2012 | SK Hynix Inc.; SK HYNIX INC | Semiconductor memory device mapping external address as internal address wherein internal addresses of spare cells of two blocks differ by upper most bit and internal addresses of main cells of two blocks differ by upper most bit and the internal addresses of main cell and spare cell of each block differ by one bit and operating method for the same |
9531646, | Dec 07 2009 | Altera Corporation | Multi-protocol configurable transceiver including configurable deskew in an integrated circuit |
9727504, | Feb 29 2012 | Mitsubishi Electric Corporation | Data transfer apparatus, data transfer method, and data transfer program |
Patent | Priority | Assignee | Title |
5563891, | Sep 05 1995 | Industrial Technology Research Institute | Waiting time jitter reduction by synchronizer stuffing threshold modulation |
6721825, | May 28 1999 | Mitel Networks Corporation | Method to control data reception buffers for packetized voice channels |
6812870, | Sep 11 2003 | XILINX, Inc. | Enhanced 8b/10b encoding/decoding and applications thereof |
6870390, | Sep 11 2003 | XILINX, Inc. | Tx line driver with common mode idle state and selectable slew rates |
6908340, | Sep 11 2003 | XILINX, Inc. | Method and system for reducing crosstalk in a backplane |
6956442, | Sep 11 2003 | XILINX, Inc. | Ring oscillator with peaking stages |
6959015, | May 09 2001 | Crest Microsystems | Method and apparatus for aligning multiple data streams and matching transmission rates of multiple data channels |
6975132, | Sep 11 2003 | XILINX, Inc. | DAC based driver with selectable pre-emphasis signal levels |
6976102, | Sep 11 2003 | XILINX, Inc. | Integrated circuit with auto negotiation |
6995618, | Sep 11 2003 | XILINX, Inc. | VCO feedback loop to reduce phase noise |
7015838, | Sep 11 2003 | XILINX, Inc. | Programmable serializing data path |
7047457, | Sep 11 2003 | XILINX, Inc. | Testing of a multi-gigabit transceiver |
7073148, | Sep 11 2003 | XILINX, Inc. | Antenna violation correction in high-density integrated circuits |
7092689, | Sep 11 2003 | Xilinx Inc. | Charge pump having sampling point adjustment |
7099426, | Sep 03 2002 | XILINX, Inc. | Flexible channel bonding and clock correction operations on a multi-block data path |
7109809, | Sep 11 2003 | XILINX, Inc. | Method and circuit for reducing VCO noise |
7187709, | Mar 01 2002 | Xilinx, Inc | High speed configurable transceiver architecture |
7188283, | Sep 11 2003 | XILINX, Inc. | Communication signal testing with a programmable logic device |
7224951, | Sep 11 2003 | XILINX, Inc. | PMA RX in coarse loop for high speed sampling |
7280590, | Sep 11 2003 | XILINX, Inc. | Receiver termination network and application thereof |
7295639, | Jul 18 2003 | XILINX, Inc. | Distributed adaptive channel bonding control for improved tolerance of inter-channel skew |
7313176, | Sep 11 2003 | XILINX, Inc. | Programmable on chip regulators with bypass |
7362864, | Sep 11 2003 | XILINX, Inc. | Framing of transmit encoded data and linear feedback shifting |
7382823, | Feb 22 2002 | XILINX, Inc.; Xilinx, Inc | Channel bonding control logic architecture |
Executed on | Assignor | Assignee | Conveyance | Frame | Reel | Doc |
Sep 04 2003 | CORY , WARREN E | Xilinx, Inc | ASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS | 014496 | /0928 | |
Sep 09 2003 | KRYZAK, JOSEPH NEIL | Xilinx, Inc | ASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS | 014496 | /0928 | |
Sep 11 2003 | XILINX, Inc. | (assignment on the face of the patent) | / |
Date | Maintenance Fee Events |
Oct 15 2012 | M1551: Payment of Maintenance Fee, 4th Year, Large Entity. |
Oct 14 2016 | M1552: Payment of Maintenance Fee, 8th Year, Large Entity. |
Oct 14 2020 | M1553: Payment of Maintenance Fee, 12th Year, Large Entity. |
Date | Maintenance Schedule |
Apr 14 2012 | 4 years fee payment window open |
Oct 14 2012 | 6 months grace period start (w surcharge) |
Apr 14 2013 | patent expiry (for year 4) |
Apr 14 2015 | 2 years to revive unintentionally abandoned end. (for year 4) |
Apr 14 2016 | 8 years fee payment window open |
Oct 14 2016 | 6 months grace period start (w surcharge) |
Apr 14 2017 | patent expiry (for year 8) |
Apr 14 2019 | 2 years to revive unintentionally abandoned end. (for year 8) |
Apr 14 2020 | 12 years fee payment window open |
Oct 14 2020 | 6 months grace period start (w surcharge) |
Apr 14 2021 | patent expiry (for year 12) |
Apr 14 2023 | 2 years to revive unintentionally abandoned end. (for year 12) |